Download 12110/99

Lynsey Wolter
12110/99
The Case of Predicates: Questions of Control and Binding
The case marking of secondary predicates in modern Russian presents interesting problems for a
Government and Binding account of Case in Russian.' Secondary predicates in Russian are
adjectives which must be case-marked. Unlike nouns, they do not move to receive Case by
government; unlike many other adjectives, they do not necessarily inherit Case from a noun. In
order to extend OB to account for the case-marking of secondary predicates in Russian, we must
allow the possibility of APs agreeing in case with more distant NPs and for the possibility of APs
recei ving default case-marking. In this paper, I will show that the statement of these mechanisms
is far from straightforward. In fact, none of the analyses that I present are entirely successful.
{The most successful analyses of the data suggest that we may need to revise the theory.}
In sections 1 and 2 I will present basic case marking in Russian and the case marking of
secondary predicates in Russian, respectively. Sections 3, 4 and 5 will discuss three types of
analyses of the case marking of secondary predicates. The analyses in section 3 are the most
conservative; section 4 presents an argument for case-marked PRO, while section 5 addresses a
possible connection between the secondary predicates and anaphora.
1. Case Marking in Russian
Every NP and AP in Russian must be marked with one of six (morphological) cases. Typically,
subjects are nominative. direct objects are accusative, and indirect/oblique objects are dative, as
shown in (1-3) below.
shown in (1-3) below .
. I am gratcfulto thc mcmbers of my thesis committce, TcJ FernalJ anJ Donna 10 Napoli; to Kari Swingle , \\ ho
mighl as wcll havc becn a thirJ member of my ~ommillce; to Sara Coc anJ Emily Manetta for hclpful ~omments on
carlier vcrsions (lIthe thcsis; anJ cspcl:ially tll m) patient ini"unnanls. V~,i1i) Dostoinc\ . Gena Kalscnclcnboigcn,
DlIllilriy Levin. Vyal:hcsbs Lukin anJ Olga RoslaFX=hova. All mistakcs arc minc.
2
(1)
Ivan
Ivan(N)
'Ivan died'
(2)
Ivan
eital
Ivan(N)
read(m.sg)
'I van read althe book'
(3)
I van
dal
knigu
Ivan(N)
gave(m.sg)
book(A)
'Ivan gave althe book to Irina'
umer l
died( m.sg) 2
knigu
book(A)
Irine
Irina(D)
This is hardly an unusual pattern of case marking; as with numerous other languages, it is
generally accepted that case is assigned in sentences like (1-3) by structural case assignment.
I[+agr] assigns nominative case to its specifier; V assigns accusative and dative case to the
sisters of V and V', respectively.
Particular V sand Ps can also assign idiosyncratic cases to their complements in Russian;
this is known as lexical or "quirky" case, and overrides any structural case assignment that might
otherwise occur. Example (4) below shows a verb which assigns genitive to its complement,
while in (5), the verb assigns instrumental to its complement. Example (6) shows one of a small
class of verbs with no arguments. Example (7) shows one of a small class of verbs with one
argument that takes accusative case. Whether this argument should be analyzed as a subject or an
object is unclear, but I will not address that issue here. Prepositions always assign lexical case to
their complements; they may assign any of the five cases we have seen so far except nominative.
Prepositions may also assign a sixth case, known as prepositional case. Prepositions with noun
complements in various cases are shown in (8).
(4)
my
vsegda
dostigaem
we(N) always
achieve( 1pI)
'We always -achieve good results'
(5)
Ivan
boleet
grippom
(van(N)
be-ill(3sg)
flu(l)
'We always achieve good results'
(5)
Ivan
boleet
(van(N)
be-ill(3sg)
'Ivan is ill with the flu'
xorosix
good(G)
resultatov
results(G)
grippom
flu(l)
(examples 4.5 from Davis & Oprendek 1972, p. 102, 196)
, The following abbre\iations will be used in the glosses in this paper: N = Nominativc, A = Accusativc,
D = Dativc, G = Geniti\c, I = Instrumental, P = Prepositional; m = masculine, f = feminine, n = ncuter;
sg = singular, pi = pluml; 1.2,3 = first, second, third person, rcspectivcly; INF = infiniti\c.
, Note that Russian vcrbs agrce With their subJecL~ in gender and number in the past tense. In the present anu futurc
tenses, \crbs agree with thcir subjccts in person and number.
3
(6)
temneet
get-dark(3sg)
'It is getting dark'
(7)
menja tosnit
me(A) be-nauseous(3sg)
'I feel ill'
(8)
(a)
(b)
(c)
(d)
(e)
posle obeda
after lunch(G)
cerez
cas
in/after
hour(A)
'in an hour'
s
bratom
with brother (I)
'with my brother'
televizoru
po
on
television(D)
0
muzyke
about music(P)
There are many complications to this basic picture, of course. Case is often correlated
with semantics. Instruments, not surprisingly, take instrumental case. Likewise, adjunct
modifiers which are time expressions (i.e. which form a semantic class) all take accusative case.
The examples in this section have all been simple declarative sentences, and different types of
constructions show different patterns of case marking. For instance, the agent argument in a
passive construction (i.e. the subject of the corresponding declarative construction) takes
instrumental case. Each of these quirks is worthy of its own paper, but 1 will not pursue them
here. Instead, I will focus on the assignment of case to predicative adjectives.
2. Secondary Predication and Case
In Russian, an adjective can function as a secondary predicate in a sentence. According to the
literature on this issue, most adjectives follow the pattern of case marking shown in (9-15)
In Russian, an adjective can function as a secondary predicate in a sentence. According to the
literature on this issue, most adjectives follow the pattern of case marking shown in (9-15)
below.) (Similar examples are cited in Greenberg and Franks 1991). I have used the same verb,
prijti 'come,' in most sentences to emphasize the fact that the form of the secondary predicate
depends on the structure of the sentence, not on the choice of the verb that is closest to the
At least one of my informants finds some additional cases acceptable on some of the prcdicativc adjectives in (915); see section 2.1.
3
4
secondary predicate.
(9)
Ivan
prisel
Ivan(N)
came(m.sg)
'I van came drunk'
pjanyj/pjanym-l
drunk(NII)
(10)
Ivan
naSel
Ivan(N)
found(m.sg)
'Ivan found Olegi drunki'
OJega
Oleg(A)
(1t)
vaZno
ne
prijti
not
come(lNF)
important
'It's important to not come drunk'
(12)
mne vaZno
ne
prijti
meeD) important
not
come(INF)
'It's important for me to not come drunk'
(13)
Ivan
poprosil
Olega
ne
Ivan(N)
asked(m.sg) Oleg(A)
not
'Ivan asked Oleg to not come drunk' (ambiguous)
(14)
Ivan
xocet
Ivan(N)
wants(3sg)
'Ivan wants to come drunk'
(15)
Ivan
pil
tol'ko vodu,
ctoby ne
Ivan(N)
drank(m.sg) only water(A), in-order-to not
'Ivan drank only water, in order to not come drunk.'
*pjanogo/pjanym
drunk(*A/I)
pjanym
drunk(l)
prijti
come(INF)
pjanym
drunk(l)
prijti
come(INF)
pjanym
drunk(I)
pjanym
drunk(I)
prijti
pJanym
come(INF) drunk(I)
It appears that the typical predicative adjective appears in a default case (instrumental) and can
also optionally agree in case with its argument when the argument is a nominative subject in the
same clause. Note that secondary predicates always agree in number and gender with their
logical arguments, regardless of the syntactic position of the arguments. Clearly, case agreement
and number/gender agreement are governed by different principles; 1 will discuss only case
agreement in this paper.
Two adjectives, odin 'alone' and sum 'oneself' show a different pattern of case marking
agreement in this paper.
Two adjectives, odin 'alone' and sum 'oneself' show a different pattern of case marking
when they are used as predicates. Although Russian grammarians traditionally analyzed the case
• Out of context, the choice of ca..<;e in sentences like this docs not appear to correlate to any difference in the
meaning of the sentence. Howe\cr, Timberlakc (1986) shows lhal, in L"onlext, there is a subtle JilTerence in meaning
between the two forms. According to Timberlake. "The lwo ca...,es ditTer ... in that the nominative signals that the
state docs nut represent a dcparture from the expected slate uf affairs, thc instrumental that the Slate docs represent a
departure" ( I<)XO: 137).
5
of these adjectives as a historical relic, Comrie (1974) argues that the pattern of case marking is
still productive in Russian syntax. Comrie points out that in Old Russian and early modem
Russian this pattern of case marking was far more productive, applying to adjectives as diverse
in meaning as pust 'empty,' Ziv 'alive' and posa:,en 'impaled.' Since this pattern was once
productive, a theory of grammar should be able to account for the pattern in a principled matter.
Other American linguists have followed Comrie's lead on this issue.
The exceptional secondary predicates agree in case with the argument they are predicated
of, so long as it is in the same clause. In (16) and (18) below the predicate agrees in case with the
subject, while in (17) the predicate agrees in case with the direct object. Example (19)
demonstrates that case agreement also occurs within an embedded (tensed) clause, as we expect.
(16)
Ivan
prisel
Ivan(N)
came(m.sg)
'Ivan came alone'
odins
alone(N)
(17)
naSel
Ivan
Ivan(N)
found(m.sg)
'Ivan found Olegi alonej'
Olega
Oleg(A)
odnogo
alone(A)
(18)
Ivan
naSel
Ivan(N)
found(m.sg)
'Ivani found Oleg alonei'
Olega
Oleg(A)
odin
alone(N)
(19)
Ivan
xocet,
ctoby Oleg
Ivan(N)
wants(3.sg), so-that Oleg(N)
'Ivan wants Olegi to come alonei'
prisel
came(m.sg)
odin
alone(N)
When an exceptional secondary predicate is in an infinitival clause, it often takes dative case,
regardless of the case marking of the overt NPs in the sentence. The (apparently) non-agreeing
dative case is shown in (20-22) below.
(20)
vazno
prijti
jrnnnrt<lnt
('nml"(fNF)
odnomu
alone(D)
dative case is shown in (20-22) below.
(20)
vazno
prijti
odnomu
important
come([NF)
aione(D)
'It's important to come alone'
(21)
mne vazno
prijti
odnomu
me(D) important
come(lNF)
alone(D)
'It's important for me to come alone'
'For simplicity, [ will use odilllhroughout this discussion. The adjective sam follows the same pattern.
6
(22)
Ja
poprosil
Ivana
I(N) asked(m.sg) Ivan(A)
•I asked Ivan i to come alonei'
prijti
come(INF)
odnomu
alone(D)
However, an exceptional secondary predicate in an infinitival clause sometimes agrees with the
nominative subject of the higher clause, as shown in (23) and (24):
(23)
ja
poprosil
Ivana
prijti
odin
I(N) asked(m.sg) Ivan(A)
come(lNF)
alone(N)
'Ii asked Ivan to come alonei' (EI 'I asked Ivan if I could come alone')
(24)
Ja
xoCu
Zit'
I(N) want(lsg)
live(lNF)
'I want to live alone'
odin
alone(N)
Interestingly, the presence of an overt complementizer appears to block agreement between the
secondary predicate in an infinitival clause and the nominative subject in a higher clause, as
shown in (25):
(25)
Ja
pereexal,
ctoby
I(N) moved(m.sg), in-order-to
'I moved in order to live alone'
Zit'
live(lNF)
odnomu
alone(D)
My discussion of the exceptional secondary predicates will necessarily include assumptions
about the structure of infinitival clauses in Russian. In keeping with standard GB theory, I
assume that every clause has a subject. The infinitival clauses in (20-25) obviously do not have
overt SUbjects. According to GB, the subject of each infinitival clause must be trace or PRO.
That is, either the subject of the infinitival has moved to subject position in the higher clause, or
it is PRO, an abstract noun phrase that can be assigned a theta-role by the main verb, but is not
governed or assigned case. The interpretation of PRO is controlled by a noun phrase in the
matrix clause, and in the absence of a controller takes an arbitrary interpretation.
I also assume the Theta Criterion, which states that a predicate must assign a theta-role to
-~~~.~I" :.~ np~ .. ~~~.~
matrIX clause, and
In
n~rI thnt ~n"l,. <>rn"rn'>r1 [ nAllp
nh ..",,,p fn""t ..prp;vp
py~('tlv
the absence ot a control er taKes an aroltrary mterpretatlon.
onp
thpt~
rolp
.
I also assume the Theta Criterion, which states that a predicate must assign a theta-role to
each of its arguments, and that each argument noun phrase must receive exactly one theta role.
Clearly, an embedded subject - which is an argument - will receive a theta-role from the
predicate of the embedded clause, and therefore can only move to the subject position of the
higher clause if the predicate of the higher clause does not have an external argument. For
convenience, I will refer to a predicate which does not have an external argument, and allows
7
NP-movement to its subject position, by the traditional teon hraising predicate." We can see that
the higher predicates of (22-25) have external arguments, and are therefore not raising
predicates; the embedded subject in each of these must be PRO. On the other hand, it appears
that va:no 'important', in (20-21), does not have an external argument, since 'me' is optional in
the examples. If vaino had an external argument, then we would expect (20), in which there is
no NP to receive a theta-role from va:.no, to be ungrammatical. Therefore, in (20-21) it is
possible that the subject of the embedded clause moved to the subject position of the higher
clause. However, evidence from idiom chunks indicates that such movement has not taken place.
One of the classic arguments that seem is a raising predicate is that an idiomatic reading
is possible with a sentence like "The cat seems to be out of the bag." That is, the apparent subject
of "seem" is licensed by the lower clause. If we try similar constructions in Russian, we find that
va:no 'important' does not act like a raising predicate in this regard:
(26)
(27)
medved'
na
uxo
nastupil
bear(N)
on
ear
stepped(m.sg)
'X is tone deaf'
vaino
na
uxo
nastupit'
medved'u
bear(D)
important
on
ear
step(INF)
literal reading only: 'It is important to a bear to step on X's ear.'
In fact, not even
~et'sja
'seem' acts like a raising verb with idiom chunks:
8
(28)
*medved'
kazetsja
bear(N)
seems(3sg)
'X seems to be tone deaf'
na
on
uxo
ear
nastupit,6
step(INF)
Arguably, there are no raising constructions in Russian 7 • Therefore we need only to account for
the distribution of predicative adjectives with respect to PRO.
The case of predicative adjectives in infinitival clauses - that is, secondary predicates
whose argument is PRO - appears to depend on the controller of PRO. If PRO is uncontrolled, as
in (20), or controlled by something other than a nominative subject, as in (21-22), the secondary
predicate has dative case. If PRO is controlled by a nominative subject, as in (23-24), the
secondary predicate has nominative case; as we saw above, this long-distance case agreement is
blocked by an overt complementizer, as in (25). Incidentally, the apparently non-agreeing dative
predicative adjective (seen in examples (20-22) and (25) has been called the Second Dative, a
convention I will adopt here.
2.1. Additional Data
My infonnants found a number of fonns to be fully or marginally acceptable in addition to the
fonns that are usually cited in the literature. Specifically, many younger informants accept the
e Example (28) is grammatical if the verb 'step' is tensed:
na
uxo
nastupil
medved' kaietsja
bear(N) seems(3sg)
on
ear
stepped(m.sg)
'It seems that X is tone-deaf.'
Here, medved' 'bear' receives Case from the tensed verb nasrupil 'stepped,' so there is no motivation for 'bear' to
move. The same pattern holds for non-idiomatic sentences: 'secms' can appear with a tensed but not with an
untensed verb, as shown in (ii-iv). Note that in (ii), the only grammatical example, 'seems' docs not agree with the
nominative NP, but thc lower \'erb does, which is consistent with the claim that the nominative NP is the subject of
the lower clause.
v
uni vcr.;i tcte
ueatsja
(i i)
om
kal.etsj'l
univcrsiLy(P)
aL
sLudy(3pl)
thcy(N) sccms(3sg)
'It seems that they study at the univcrsity .'
(i)
117~~lh~i:'I
iii i\
~/"\ni
(i i)
ucatsja
om
kal.ctsja
sLudy(3pl)
thcy(N) sccms(3sg)
'It scems that they study at the univcrsi ty.'
L.-!"l'711IcI :,
v"
at
Iln i \':f""r~;t('otl'"
univcmtete
univcrsiLy(P)
\'
umversi tete
ucatsJU
*nni
kal.uL".ia
uni veri sty( P)
at
study(3pl)
they(N) seem(3pl)
(i\)
*oni
kaJutsja
ucit'sja
v
universitete
they(N) secm(3pl)
study(lNF)
at
univcrsity(P)
It appears that ka:efsja 'seems' is not a raising verb. In these sentences, it is possible that 'seems' is a parenthetical,
or alternatively that the lower subject has been fronted into a topic positivc.
- Schein (1%1) argues that subjects of small clauses r.lise in Russian. For the purposes of this discussion, I will
disregard raising from small clauses.
(iii)
9
default instrumental with nearly all instances of the exceptional predicative adjectives. They also
appear to allow case agreement in a wider range of environments. However, much of this
apparent case agreement is due to homophony between true secondary predicates and floated
quantifiers.
2.1.1. The Default Instrumental
Recall that the instrumental case is the "default case" for regular predicative adjectives. Several
of my younger informants almost always allow this default case with the exceptional
predic~tive
adjectives as well, as shown in (29-34) below; the instrumental was rejected only in (29). Note
that these sentences are acceptable in addition to the standard paradigm given in Section 2.
Generally, the forms in (29-34) are considered more colloquial than those in the standard
paradigm.
odinl*odnim
alone(N/*I)
(29)
Zivy
Ja
I(N) live(lsg)
'I live alone'
(30)
ego
Ja
naSel
I(N) found(m.sg) him(A)
'I found himi alonei'
(31)
(mne) vaZno
zit'
odnim
meeD) important
live(INF)
alone(l)
'It's important (for me) to live alone'
(32)
Ja
poprosil
Ivana
I(N) asked(m.sg) Ivan(A)
'I asked Ivani to live alonej'
(33)
Ja
I(N)
'I
xocu
want( lsg)
w~nt to
livf'
Zit'
live(lNF)
odnim
alone(l)
Zit'
live([NF)
odnim
alone(1)
odnim
alorie(l)
~Ione'
(33)
Zit'
xocu
Ja
live(lNF)
I(N) want( lsg)
'I want to live alone'
odnim
alorie(l)
(34)
Ja
pereexal,
ctoby
leN) moved(m.sg). in-order-to
'I moved in order to live alone'
zit'
live(lNF)
odnim
alone(1)
In (29), the example in which the instrumental is rejected, the secondary predicate's argument is
10
the nominative subject of the same (simple) clause. This is the same environment in which
regular predicative adjectives optionally agree with their subjects. There is clearly something
about local subjects which is conducive to case agreement. Other than the peculiarity of (29), we
can say that for younger speakers the "default instrumental case" has generalized from regular to
exceptional predicative adjectives.
2.1.2. More on Case Agreement
My informants often appear to allow or require case agreement between a secondary predicate
and its argument when that argument is the direct object of the primary predicate. One speaker
found (35) acceptable:
(35)
ja
naSel
ego
I(N) found(m.sg) him(A)
'I found himi cheerfuli'
veselogo
cheerful(A)
In (35), a regular predicative adjective agrees in case with an accusative direct object in the same
clause. We only need a minor revision of the descriptive generalization to account for this. We
can now say that, for some speakers, regular predicative adjectives agree in case with any
argument in their own clause - not just a nominative argument. The default instrumental case is
still always acceptable for regular predicative adjectives. The next section will discuss another
apparent occurrence of case agreement between a secondary predicate and its argument;
however, I will argue that these "secondary predicates" are, in fact, floated quantifiers.
2.1.3 . Roated Quantifiers
An intriguing pattern of results from my informants is shown in (36-38):
(6)
la
velel
emu
odnomu
Zit'
An intriguing pattern of results from my informants is shown in (36-38):
(36)
(37)
Ja
velel
emu
l(N) ordered(m.sg) him(D)
'I ordered himi to live alonei'
Ja
ugovoril
ego
I(N) persuaded(m.sg)
him(A)
' I persuaded himi to live alonej'
Zit'
live(lNF)
odnomu
alone(D)
zit'
live(lNF)
odnogo
alone(A)
11
(38)
Ja
poprosil
ego
I(N) asked(m.sg) him(A)
'I asked himi to live alonei'
Zit'
live(lNF)
?odnogol?odnomu
alone(?AI?D)
At first glance, these sentences appear to show more long-distance case agreement. Each of these
sentences contains an embedded infinitival clause; the secondary predicate's argument in each
sentence is PRO, which is controlled by a direct or indirect object. According to the literature,
the secondary predicate should be dative in each of these sentences. But notice that my
informants consistently prefer the accusative case for the secondary predicate in (37). Example
(38) is more problematic; both accusative and dative are marginally acceptable.
However, not all speakers accept (36-38). Speakers who do accept (36-38) sometimes
indicate that the secondary predicate is in a strange position, and that they would prefer to put it
earlier in the sentence, next to the antecedent. But when odin and sam occur next to a noun, they
function as quantifiers rather than predicates. This is especially clear with odin. As Comrie
(1974) points out, odin translates as 'only' when it is used as a quantifier and 'alone' when it is
used as a secondary predicate, as shown in (39-40) below. Note that this difference in meaning is
similar to the English 'I, alone, was sitting' versus 'I was sitting alone.'
(39)[=10]
odin
ja
only(N)
I(N)
'Only I was sitting'
sidel
sat(m.sg)
(40)[=11 ]
Ja
sidel
I(N) sat(m.sg)
'I was sitting alone'
odin
alone(N)
Therefore, we can conclude that odin in (36-38) is an (incorrectly translated) floated quantifier,
rather than a predicate H • I will not consider the forms in (36-38) in the remainder of the paper, but
will assume that, in each of these sentences, a true exceptional secondary predicate must occur in
will assume that, in each of these sentences, a true exceptional secondary predicate must occur in
the dati ve case.
• Schein (19~Q) apparenLly reached a similar conclusion.
12
2.2. Summary
The tables below summarize the case marking of predicati ve adjectives. In each schema, the
predicative adjective and its argument are boldfaced, and relevant examples are cited to the right.
Note that in (41c), (42d) and (43d) I have not indicated the grammatical function of the dative
noun phrase at the beginning of the sentence. The grammatical function of these dative NPs is
the subject of some controversy; I will address this issue in section 4.1.2.
(41) Schematic table of case-marking of regular predicative adjectives
structure
cf.
(a)
SUBJ(N)
V
(OB1)
PRED(N/I)
(9)
(b)
SUB1(N)
V
OBJ(A)
PRED(I)9
(10)
(c)
(NP[DDi
ADV
[PROi V
PRED(I)]
(11,12)
(d)
SUB1(N)i
V
(OB1) [PROi
(e)
SUB1(N)i
V
(OB1) [C
PROi
V
PRED(I)]
(15)
(f)
SUBJ(N)
V
OBJ (A)i
[PROi
V
PRED(I)]
(13)
V
PRED(I)]
(13,14)
(42) Schematic table of case-marking of exceptional predicative adjectives for 2 informants
cf.
structure
(a)
SUBJ(N)
V
(OBJ)
PRED(N)
(16,18)
(b)
SUBJ(N)
V
OBJ(A)
PRED(A)
(17)
(c)
SUBJ(N)
V
LC
(d)
(NPlDj)j
ADV
[PROi V
(e)
SUBJ(N)i
V
(OBJ) [PROi
(0
SUBJ(N)i
V
(OBJ)_ IC
(e)
SUBJ(N)i
V
(OBJ) [PROi
(0
SUBJ(N)i
V
(OBJ) [C
PROi
V
PRED(D»)
(25)
(g)
SUBJ(N)
V
OB] (D/A)i
lPROi
V
PRED(D)J
(22)
V
OBJ(A)
PRED(A/l)
d. (35)
.) Or, for one informant:
SUBJ(N)
SUBJ(N)
V
PRED(N)J
(20,21 )
PRED(D)J
V
PRED(N)J
V
PROi
V
(19)
PRED(D)I
PRED(N)]
(23,i4)
(25)
(23,i4)
13
(43) Schematic table of case-marking of exceptional predicative adjectives for 3 informants
structure
cf.
(a)
SUBJ(N)
V
(OB1)
PRED(N)
(16,]8,29)
(b)
SUBJ(N)
V
OBJ(A)
PRED(AII)
(17,30)
(c)
SUB1(N)
V
[C
(d)
(NP[DDi
ADV
[PROi V
(e)
SUB1(N)i
V
(OB1) [PROi
(f)
SUB1(N)j
V
(OB1) [C
PROi
V
PRED(D/I)]
(25,34)
(g)
SUB1(N)
V
OBl (D/AH
[PROi
V
PRED(D/I)]
(22,32)
SUBJ(N)
V
PRED(N)]
PRED(D/I)]
V
(19)
(20,21,30,31 )
PRED(N/I)]
(23,24,33)
In (41), we see that regular predicative adjectives are always acceptable in the default
instrumental case. These predicative adjectives also optionally agree with their argument when
that argument is the nominative subject of the same clause. One infonnant has apparently
extended this case agreement, and accepts predicative adjectives that agree in case with their
argument even when the argument is a direct object in the same clause.
(42) shows the case marking of exceptional secondary predicates as given in the literature
and by two of my infonnants. Note that in (42a-b) an exceptional secondary predicate must agree
with its argument if they are both in the same clause. When the exceptional secondary predicates
occur in infinitival clauses, their case marking is more complicated. In (42e) we see that when
PRO is controlled by a nominative subject and there is no intervening -complementizer, the
secondary predicate takes nominative case. (42f) shows that when PRO is controlled by a
nominative subject but there is an overt complementizer, the secondary predicate takes dative
case. The secondary predicate also takes dative case when PRO is uncontrolled or controlled by a
direct or oblique object, as in (42d) and (42g).
case.
(4~ \
shows the
cas~
markin!!
of exceotional secondarY
predicates as oiven by three of "J
myal.su ldl\.C;~ UCllrVC:: \...C1:l1fV
l -,,\oJ . ~
I ne-sec(jno-ary-preUll.:are
V\' U'I;.II
UI.I"",VIl\,IVIl',,U VI
'"''-'''\,1_.1 .... -
direct or oblique object, as in (42d) and (42g).
(43) shows the case marking of exceptional secondary predicates as gi ven by three of my
informants. Note that (42) is nearly identical to (43): these informants accept all of the standard
forms. However, they also accept the default instrumental case with nearly all instances of
exceptional secondary predicates. The default instrumental case is only rejected when the
secondary predicate'S argument is a nominative subject and both occur in the same clause.
14
In the following several sections I will explore different analyses which might account
for the case marking of exceptional predicative adjectives. A successful analysis must be able to
account for the instances of case agreement, as well as for the appearance of the Second Dative
(the non-agreeing dative case in, e.g, (42f-g)). We prefer an analysis that includes as little
revision of standard theory as possible. However, as I show in section 3, several analyses that do
not require revisions of the theory are unsuccessful. In section 4 I present an analysis that
requires PRO to be case-marked. In section 5 I outline a different analysis, based on Binding
Theory.
3. Initial Accounts: Structural Case, Default Case and Restructuring
3.1 Accounting for the Second Dative
In this section I will consider two ways to account for the Second Dative that do not require PRO
to be case-marked. Let us assume, for the sake of argument, that the Second Dative occurs only
when long-distance case agreement is blocked, and that we can account for the blockage of longdistance case agreement. This leaves us the somewhat more manageable task of accounting for
just the occurrence of the (apparently) non-agreeing dative case on secondary predicates.
Linguists have proposed several different analyses of the Second Dative within a Government
and Binding framework. Some argue that the Second Dative arises when a structural case
assignment rule - motivated by other data - applies to secondary predicates. Another possibility
is that the Second Dative is a default case, similar (but not identical) to the default instrumental
that we saw with the regular predicative adjectives. I will explore each of these proposals in
more detail here.
3.1.1. Greenberg & Franks' (1991) Structural Case Assignment
3.1.1. Greenberg & Franks' (1991) Structural Case Assignment
Greenberg and Franks ( 1991) argue that the Second Dative is the result of a rule of structural
case assignment. This rule is motivated by the occurrence of the dative case on argument noun
phrases, but can apply elsewhere in the absence of other case-assigning mechanisms. The rule is
always available to secondary predicates, but can be overridden by case agreement. just as
15
structural case assignment to nominal arguments can be overridden by lexical case assignment.
Greenberg and Franks (1991) propose that dati ve case is structurally assigned to any (NP
or AP) sister of X'. Given the standard assumption that indirect/oblique objects are sisters of V'
(see 44), this rule accounts for the assignment of dative case to indirect/oblique objects.
(44)
VI
~
V
l'Jp ~
Vlf.6V 1 OB:;JE& T
Furthennore, the rule accounts for the so-called "dative subjects." These are dative NPs which
appear in canonical subject position in tenns of linear word order, but with which the verb does
not agree. According to Greenberg and Franks' (1991) analysis, these subjects - sisters of I' are not assigned nominative case by I[+agr] and dative case can therefore be assigned. This is
shown in (45).
IP
(45)
~
\l
/)ATH/~ SIl8'jl;cT" ~
NP
\'
_ r,' _ _ __ __11
Finally, Greenberg and Franks (1991) argue that predicative adjectives are sisters of (' as well,
by ternary branching, as shown in (46a) below. Franks (1995) pursues essentially the same
analysis, but argues that predicative adjectives are Chomsky-adjoined to 1', as shown in (46b)
below. Given either structure, predicative adjectives are assigned dative case by the structural
case assignment rule. Thus, one structural case assignment rule accounts for several instances of
16
case assignment, and an analysis of the Second Dative requires no further additions to the theory.
(46a)
(46b)
If
~U m-~I -">
/I~
NP
I ' AP
6
.$V &Jl;:C T -')0
IP
NP
~
\ .-
~
f
5ecotJDA F?y
I'-
P~eDlcAr6"
~
Ape--
~C()Nf)Af!.Y
PF.(;/) I Cit
re
Unfortunately, Greenberg and Franks' case assignment rule is far too productive. Note
that PRO is also a sister of 1', and of course no standard case-assignment mechanism will apply
to PRO. By Greenberg and Franks' account, PRO should also be assigned dative case. But this
violates our standard assumption that PRO is ungoverned and does not receive case. Greenberg
and Franks (1991) do recognize this problem, but are forced to stipulate that their caseassignment rule is governed by the tense feature of I when it applies to arguments, but not when
it applies to non-arguments. That is, the rule can apply to non-arguments at any time, but to
arguments only in tensed clauses. Although arguments and non-arguments certainly do behave
differently with respect to case assignment, there is no independent motivation for the claim that
a rule of structural case assignment is more general when it applies to non-arguments than when
it applies to arguments. Alternative stipulations (e.g. that the dative rule simply does not apply to
PRO) are equally unsatisfying. In fact, it is tempting to take Greenberg and Franks' (1991)
analysis as evidence that PRO is case-marked in Russian. I will discuss this possibility in section
4 below.
3.1.2. Dative as an Exceptional Default
I .p.tlJ".now hrif"f1v ('on"irtf'r thE' nl''i"ihilitv
3.1.L..
vattve as an t:xcepuonal ueIault
th~t
thp s.P('onrt nMivp i" ~ kinrt of rt .. f<l,,1t ,,<1.,1' for
Let us now briefly consider the possibility that the Second Dative is a kind of default case for
exceptional predicative adjectives, similar to the instrumental default case of regular predicative
adjectives. Clearly, the Second Dative cannot be exactly the same kind of default as the
instrumental default. As we have seen. the instrumental default is always acceptable with regular
predicative adjectives, even when case agreement is allowed:
17
(47)
Ivan
Ivan(N)
prisel
came(m.sg)
pjanyjlpjanym
drunk(N!I)
The Second Dative, on the other hand, is not accepted when case agreement is possible.
Therefore, we cannot simply state that the exceptional secondary predicates have an exceptional
default case. However, a more plausible alternative remains. Schein (1982) argues that dative is
an "elsewhere" case for exceptional secondary predicates. He also argues that the Second Dative
must occur when an exceptional secondary predicate does not agree in case with a nominative
antecedent, a generalization that is contradicted by data such as (17), repeated here as (48):
(48)
Ivan
naSel
Ivan(N)
found(m.sg)
'Ivan found Olegi alonei'
Olega
Oleg(A)
odnogo
alone(A)
Nevertheless, if we can account for all occurrences of case agreement in the data without
introducing major revisions to the theory, it may be sufficient to stipulate that the dative is an
"~lsewhere"
case that occurs on secondary predicates if and only if case agreement is not
possible.
3.2 Accounting for Long-Distance Case Agreement: Restructuring
In this section, I will explore one possible account for the long-distance case agreement in
sentences like (23) and (24), repeated here as (49) and (50).
(49)
ja
poprosil
Ivana
I(N) asked(m.sg) Ivan(A)
'I i asked Ivan to come alone(
prijti
come(lNF)
(50)
Zit'
Ja
xocu
live(INF)
I(N) want( 1sg)
'I want to live alone'
odin
alone(N)
(50)
Zit'
Ja
xocu
live(INF)
I(N) want(l sg)
'I want to live alone'
odin
alone(N)
odin
alone(N)
In section 3.2.1 I will outline evidence for a special transformation called Restructuring in
Italian. Restructuring essentially converts two verbs from adjacent clauses into a single complex
verb. thus converting a complex sentence into a single clause. The transformation was proposed
in the 1970's to account for certain complexities of Romance syntax. In section 3.2.2 I show how
18
.,
Restructuring may be able to account for the long-distance case agreement in Russian, and what
problems arise under such an analysis.
3.2.1 Restructuring in Italian
Restructuring has been proposed in order to account for a number of constructions in Italian and
other Romance languages in which elements of an embedded clause move to or affect the higher
clause. For example, in Italian, clitic complements of an embedded clause are sometimes allowed
to attach to a higher verb. In (51) below, mi, the clitic complement of 'see,' is attached to 'see,'
while in (52) the same clitic is attached to the higher verb 'want' (examples from Napoli 1981):
(51) [=1]
voleva veder-mi
(52) [=2]
mi voleva vedere
'She wanted to see me'
Researchers have argued that in a sentence like (52), the two verbs are reanalyzed as a single
complex verb, or an auxiliary and a main verb. Thus, although the clitic in (52) has apparently
"climbed" to a higher clause, there is really only one clause. In (51), on the other hand,
Restructuring has not occurred, and the clitic is placed in the lower clause. There are a number of
other arguments for a Restructuring transformation in Italian, which need not concern us here
(but see Napoli [1981] for an overview). The relevant point is that researchers have used
Restructuring in order to retain conditions on locality. Can Restructuring account for longdistance case agreement in Russian, and allow us to make the generalization that case agreement
is local?
3.2.2. Restructuring in Russian
The essence of the argument for Restructuring in Russian is that it will allow us to say that all
The essence of the argument for Restructuring in Russian is that it will allow us to say that all
case agreement is local. Let us first look at a concrete example of this analysis. Example (53)
below shows a sentence in which the secondary predicate apparently agrees with a NP in a
higher clause.
19
(53)
[PRO zit'
Ja
xocu
I(N) want(lsg)
[
live(lNF)
'I want to live alone'
odinJ
alone(N)]
The application of Restructuring to (53) will result in a single clause with a complex verb:
(54)
Zit'
J
Ja
[vxocu
I(N) want( Isg)
live(INF)
'I want to live alone'
odin
alone(N)
Notice that the argument of the predicate is now the nominative NP in the same clause. What we
would like to do is argue that Restructuring occurs injust those sentences which have apparent
long-distance agreement, and in no others.
Our original description of the distribution of long-distance case agreement was that it
occurs when the embedded clause is an infinitival, PRO is controlled by a nominative subject,
and there is no overt complementizer. As Restructuring in Italian can only apply to two verbs
that are linearly
adjac~nt,
the blocking effect of complementizers is not surprising. It is fairly
straightforward to rephrase these conditions in tenns of Restructuring:
(55)
Conditions on Restructuring in Russian
(a) Restructuring is obligatory in subject-control sentences without an overt
complementizer.
(b) Restructuring is blocked by an overt complementizer.
Unfortunately, this analysis raises two immediate problems. First, if Restructuring can only
occur when two verbs are linearly adjacent, then it should not occur in a sentence like (SO), in
which a direct object intervenes between the two verbs. A Restructuring analysis, then, cannot
account for all occurrences of long-distance case agreement in Russian.
Secondly, (55a) predicts that Restructuring should depend on the verbs in a sentence, and
not on a secondary predicate. We expect, based on (54) and (55), that (56) will be grammatical,
not.
but it is ..:>t:l,;UIIU1Y,
\JJaJ
pi t:Ull,;l~
lllal
"t:~1I
Ul,;lUIIIIC; =>IIUUIU Ul;;l'I;;IIU UII LII'"
Y " ' . U~
II.
<l ~""HI;;II"'''', allY
not on a secondary predicate. We expect, based on (54) and (55), that (56) will be grammatical,
but it is not.
(56)
*ja
I vxocu
zit"
I(N) want( Isg)
live(lNF)
. I want to live cheerfully'
veselyj
cheerful(N)
Note that in Romance languages, Restructuring does depend on the meaning of the verbs that are
reanalyzed, so lexical dependence per se should not cause us to abandon this analysis. However,
,
20
we certainly do not expect a secondary predicate, whose projections are not affected by
Restructuring, to govern the occurrence of the transformation.
4. Case-Marked PRO
[n section 3, we saw that attempts to account for the case marking of exceptional predicative
adjectives in Russian without introducing major revisions to GB theory were not entirely
successful. The structural case assignment rule proposed by Greenberg and Franks (1991) is too
productive, and should also assign dative case to PRO. In fact, it is not clear what kind of
structural case assignment rule could account for the data without assigning case to PRO, given
that the I that governs PRO also governs the Second Dative. A Restructuring account of longdistance case agreement was even less successful, as we saw in section 3.2.
At this point, we could certainly fall back on stipulation; given that we are dealing with
only two lexical items, it is not unreasonable to propose that the case marking of these items is
entirely specified in the lexicon. However, by doing this we risk missing generalizations about
case marking in Russian; I will set aside a stipulative account for now as a "last resort."
In this section, I will pursue a different analysis of the case marking of exceptional
secondary predicates, one which requires PRO to be case-marked. In section 4.1 I will discuss an
analysis of the Second Dative along these lines. In section 4.2 I discuss long-distance case
agreement. Section 4.3 is concerned with the implications of case-marked PRO.
4.1 Accounting for the Second Dative with Case-Marked PRO
4.1.1 Case-Marked PRO in lcelandic and Russian
Sigurdsson (1991) proposes that PRO is case-marked in modem lcelandic. He provides several
arguments in favor of this position, but I will summarize only the argument which is most
Sigurdsson (1991) proposes that PRO is case-marked in modem lcelandic. He provides several
arguments in favor of this position, but I will summarize only the argument which is most
relevant to the Russian data under consideration here.
Sigurdsson argues that in Icelandic, floating quantifiers and secondary predicates are
assigned case by belonging to "morphological case chains" headed by NPs. For example, the
.r
21
floating quantifiers in (57) and (58) agree in case with the subjects of the sentences lO :
(57) [=6a]
(58) [=6bl
stnikamir
komust
allir
the boys(N) got
all(Npl.m.)
'The boys all managed to get to school'
stnikana
vanta5i
alia
the boys(A) lacked
all(Apl.m.)
'The boys were all absent from school'
r
to
f
in
sk6la
school
sk61ann
the school
When the same floating quantifiers appear in embedded infinitival clauses, they appear to agree
in case with the case that each verb would assign to its subject, if PRO could be case marked.
Compare (57) and (58) to (59) and (60):
(59)[=8a]
stnikamir
the boys(N)
vonast til
hope for
[a5
to
PRO
(N)
komast
get
allir f
all(N) to
(60) [=8bJ
stnikamir
the boys(N)
sk6lannJ
the school II
vonast til
hope for
[a5
to
PRO
(A)
vanta ekki
lack not
alIa
all(A)
sk6la]
school
In
These Icelandic data suggest that we may want to allow lexical case assignment to apply to
PRO. I! By the same logic, we may be able to argue that structural case assignment applies to
PRO in Russian. Exceptional secondary predicates in Russian often agree in case with a noun
phrase elsewhere in the sentence. Whenever exceptional secondary predicates do not agree with
any overt noun phrase, they occur in a clause with PRO. Therefore, we might argue that PRO is
marked with dative case in Russian, and that exceptional secondary predicates always agree in
case with an antecedent.
Unfortunately, the Russian data are not as clear as the Icelandic data. Note 'that in the
Icelandic data, every "morphological case chain" appears to be confined to a single clause. That
is, whenever a quantifier appears in a clause with PRO, it will agree in case with PRO, assuming
-
•
11
.
,
Icelandic data, every "morphological case chain" appears to be confined to a sin-gle clause. That
is, whenever a quantifier appears in a clause with PRO, it will agree in case with PRO, assuming
that Sigurdsson' s proposal is correct. On the other hand, Russian secondary pred'i1&tes do not
always show dative case when they appear with PRO. As we have seen, there are-,s'offie instances
10 This analysis also assumes that Icelandic verbs assign "quirky" case to their subjects. For a ui~cussion of quirky
subjeCL'i, see Sigurdsson (1')')2).
I I Sentence glosses were not provided for these sentem;cs.
" Sigurusson (I')') l) also argues that PRO can be assigneu structurJ.I case in Icelandic, based on;~~~~'\'ing to do
\vith verb agreement.
:
22
in which a predicate appears to have a long-distance agreement relationship with an NP in a
higher clause. An analysis based on case-marked PRO must also provide some account for the
long-distance case agreement, a problem I will address in section 4.3.
If we argue that PRO.is case-marked, we must also explain how PRO can receive dative
case by structural case assignment. In fact, there is some motivation for this. Many researchers
have pointed out that dative NPs are the apparent subjects of an interesting construction in
Russian containing an infiniti val. As I mentioned in section 3.1.1, the "dative subjects" are
relevant to Greenberg and Franks' (1991) analysis; I will now discuss in more detail how the
"dative subjects" are relevant to an analysis involving case-marked PRO.
4.1.2. The "Dative Subject" in Russian
One of the more puzzling and controversial issues in Russian case marking is the problem of
dative NPs which, based on linear word order, appear to be in the subject position of sentences.
Examples of these "dative subjects" are given in (61) and (62) below.
(6l)
mne (bylo)
uxodit'
leave(lNF)
meeD) was(n.sg)
' I have/had to leave'
(62)
mne (bylo)
nado
meeD) was(n.sg)
necessary
'I have/had to leave'
uxodit'
leave(lNF)
Note that there is, no subject-verb agreement in these sentences - the verb, if any, takes a default
form, neuter singular. There is, however, tense, as indicated by the presence of the copula in the
past tense.
Sin~e
the copula is null in the present tense, we must rely on the past tense to show
the presence or absence of tense and agreement. Note also that there are two constructions with
"dative subjects~::j ln (61) the construction consists of a dative NP, a copula, and an infinitival. In
th~ pi-'esen~e ~ia6seii~e
or it;;ii~ean~rag;e·e~enCN~te also-ih~t\here-a/e t~o·construci,oii~ with
"dative subjects·~::jln (61) the construction consists of a dative NP, a copula, and an infinitival. In
(62) there is also an; adverb; when the adverb nudo'necessary'lJ is used, the two constructions
are semantically v.ery similar.
The grammatical function of the dative NPs in (61) and (62) is by no means clear.
Researchers have argued that they are subjects; that they are indirect objects; and that they are
I'
Nado is
m()rph(W!)~iciti ly identifiable as an advcrb, but convcntionally translated as "necessary."
23
some of each l ", • But note that if we assume that these dative NPs are true subjects, then we can
give a simple explanation of the (possible) dative case-marking of PRO. Note that, like PRO, the
"dative subjects" appear in a clause without subject-verb agreement. We assume that If+agrl
assigns nominative case; now we need only add that in Russian, If-agrl assigns dative case. Thus,
both the "dative subjects" and PRO will receive dative case.
4.2. Accounting for Long-Distance Case Agreement with Case-Marked PRO
In the previous section I presented an argument that PRO is structurally assigned dative case in
Russian. The Second Dative is then the result of local case agreement: a secondary predicate
agrees in case with PRO in the same clause. However, we have seen that secondary predicates do
not always take the dative case when they occur in infinitival clauses. Recall that when PRO is
subject-controlled and there is no overt complementizer, long-distance case agreement is
possible (cf. 42). In order to extend this analysis to cover the long-distance case agreement, we
must make one of two possible revisions: we must provide a way for the secondary predicate to
agree in case with a higher su~iect when necessary, or we must provide a way for PRO to receive
cases other than dative.
Suppose that the dative secondary predicate agrees in case with PRO and the nominative
secondary predicate agrees in case with a higher subject. This suggests that the secondary
predicate chooses the "best" antecedent from among its argument and the NPs that are coindexed
with the argument. A fully referential subject is better than PRO, but PRO - a subject - is
, ". ," \ ~q,~,"
preferable to a fully referential object. This suggests that a more detailed investigation of the data
in an optimality theoretic framework might be worthwhile in the future. An analysis involving
Optimality Theory is beyond the scope of this study, however.
On the other hand, if we want to extend our analysis without introducing'Optimality
Optimality Theory is beyond the scope of this study, however.
On the other hand, if we want to extend our analysis without introducing;Optimality
Theory, we could argue that the exceptional secondary predicates agree in
cas~\~Ythd local
argument, and that PRO actually is assigned nominative as well as dative c;se". Urider'this
14 See Bachman (1980) for a summary of the positions of Russian linguists on this issue. ~c~M) lJ~\~r argues.
working within the theory that preceded GB, that dative subjects arc, in fact, subjects. In more recent work, Preslar
( 1~5) argues that "dative subjects" are actually indirect objeCL'i, while Greenberg and Fr.lnk~ ~ IWj, ,lake the
position that some apparent "dative subjects" are subjecL-;. while others are indirect objects.
24
analysis, PRO is assigned dative case unless its controller is a nominative subject and there is no
overt complementizer. This implies that PRO inherits case from its subject controller. The
conditions on the case-marking of PRO are then as follows:
(63)
Case-marking of PRO
(a) PRO agrees in case with a subject controller.
(b) Case agreement between PRO and its controller is blocked by an overt
complementizer.
(c) Case agreement overrides structural case assignment.
We can now account for all of the case-marking of exceptional secondary predicates. PRO is
structurally assigned dative case by I[-agr]. When the secondary predicate's argument is PRO,
the secondary predIcate will agree in case with PRO. In addition, under the conditions outlined
above, the second~ry predicate agrees with a higher subject instead of PRO or PRO inherits
nominative case from its controller. Unfortunately, this successful analysis only comes at the
i. ..
j j ~... .
~
expense of disposing with the PRO Theorem. We can account for the case marking of
,,.,. J ~
•
predicative adjectives only by losing our explanation of the complementary distribution of PRO
-.
~ ,.
,:,j ~ ~
and lexical NPs. lhe next section will address this problem.
4.3. The Implication~ of Case-Marked PRO
If we allow PRO to be case-marked, and assigned case by I[ -agr], we essentially destroy the PRO
Theorem, which states that,PRO is ungoverned and not assigned Case. Therefore, PRO appears
1~
~; " ~
. ,.'
- : .. ~ ,
in exactly those positions where lexical NPs are not licensed, since lexical NPs must receive
'; )': :r-: 2 f: _
.
~
Case. In this section I have presented an argument which states that PRO is governed and
·
~
t;' · ('.
",'
j
;
,
~
assigned Case, at least in Russian and Icelandic. If we accept this argument, we must find a new
way to account for the distribution of PRO and lexical NPs.
•
,-
_;-
(~
.!. , : '. I ; .- . . ,
' • .J"
"orne res.earcher~ h:lve mane 1lr:oume8t~ ahout the Rll~sian data which are veN ~i rnil::1r to
asslgnea Lase, :n least tn KUsslan ano lcelan IC. I"t we accep[ tms argumem, we mus[ Ilno a new
way to account for the distribution of PRO and lexical NPs.
•
,-
_'
(~
",.J"
.!. . : '.! ;
,. . . ,
Some researchers have made arguments about the Russian data which are very similar to
. nJ'
J' ~,; .
.
mine but avoida1~A~ri~,~ C~se to PRO in the syntactic derivation proper. Comrie (1974) presents
an argument that is essentially the same as mine but which predates Control Theory. Clearly, an
analysis whichinv61-ve's tleleting lexical NPs from infinitivals is no longer useful. Bouchard
: ,TJ'J1Jl! ,';
(1984) proposeHMHDRO is case-marked - but only at LF. One would prefer a more principled
25
analysis than one which relegates case marking to LF, especially since case:. marking is generally
associated with PF.
Sigurdsson (1991), addressing this problem for Icelandic, proposes that lexical NPs must
be licensed by "proper head government." That is, a lexical NP must be governed by one of a list
of "proper head governors;" the list of such governors may be a parameter of universal grammar.
Crucially, I[-tense], which governs PRO, is not a proper head governor. This amounts to a
stipulation that lexical NPs cannot appear in the subject position of tenseless clauses; this is
probably the best we can do, because once we allow PRO to be governed and assigned case,
there is little else to differentiate PRO and lexical NPs.
We seem to be faced with a choice between two stipulations. If we accept the only
working analysis of the case marking of exceptional secondary predicates that we have seen so
far, then we must stipulate the distribution of PRO and lexical NPs. If we retain the PRO
Theorem, then we seem to have no way to account for the case marking of excepti6nal secondary
predicates other than by specifying it in the lexicon. This unsatisfying conclusion 'suggests that it
may be worthwhile to reexamine our assumptions. Recall that we assumed that the exceptional
case-marking pattern is productive in modem Russian, even though it only
appl~es
to two words.
In the next section, I discuss the possibility that these two words have in fact been.J~analyzed as
~.'
• f
I
anaphors in modem Russian.
5. Predicative Adjectives and Binding Theory
_ r ..,! '.,l.; v.:f ,-" ~. ~ " "
.
It may not be a coincidence that the only two exceptional predicative adjec:tives~n modem
.(
... , -...
~
.
~
Russian, odin 'alone' and sam 'self,' are semantically similar to reflexiyes.;Th~~¢'~'~~ interesting
.. ,
~ \_~ :,;'!1:~ rh
I
similarities between anaphors and exceptional predicative adjectives in Rvssian, "which I will
" '
• ,:
. f"· • ~
~ ':.
f ." ,
discuss in this section. The similarities suggest that the binding of reflexiv~s ,a ndJhe case;
similarities between anaphors and
,
(~I
:\
pI
exc~ptional pr~di~ative- adjectives i~ R~ssia~L"~hi'~h' I w'ill
"
,',
- ,"
~t' r .. ~
~
':::. .r.",
discuss in this section. The similarities suggest that the binding of reflexiv~sand , the~asemarking of exceptional secondary predicates result from the same or similar mea~~'i~rits. In this
section, ( will first discuss typological generalizations about two kinds
of anapfihrs.(I will then
,
. .. ,L::Il E:
relate the case marking of exceptional secondary predicates to the binding Qf.taftaph~~s. and argue
that the exceptional secondary predicates are, in fact, a kind of anaphor. Although the details of
the mechanism of binding of these "predicate anaphors" remains to be defined, this conclusion
26
'.
brings together a number of geper.~ ~:! zations about the case marking of exceptional secondary
predicates without contradicting standard theory .
5.1. Local and Long-Distance Anaphora
Although standard Binding Th.~ory states that an anaphor must be locally bound, numerous
researchers have noted that there are actually two kinds of anaphors, each with different
conditions on binding. Standard Binding Theory accounts for morphologically complex
anaphors; a number of recent studies have attempted to revise or extend Binding Theory to
account for the behavior of morph,qlogically simple anaphors as well. Let us first look at each
kind of anaphor in more detaij.
Morphologically
corqpl~.X anaphors, such as English himself, must be locally bound.
. .,;...
.
.
Although anaphors such as. hfin.seljare
usually bound by a subject, noun phrases with other
,
.,.:.. .: ~
grammatical functions
are~lso., ~cceptable
--
(...." , .
,
as antecedents to morphologically complex anaphors.
Thus, we have:
(64)
(65)
John saw himself. , " "
,
They showed John himselfin the mirror.
Morphologically sirrtple'(trionomorphemic) anaphors, on the other hand, are not always
locally bound. In Chinese, the monomorphemic anaphor ziji 'self' can be bound by an
antecedent at any distance, as shown in (66) below (from Cole and Sung, 1994).
(66)[=2]
Zhangsani
Zhangsan,
'Shangs~n
,
renwei Lisij zhidao Wangwuk
xihuan
, IZ th,ink Lisi
know Wangwo
line
thihksLisi knows Wangwu likes him/himself.'
zijii/jlk
self
,:- ;2 i! :'"tf~'l l'r:: :.: ~.7 . '; .. .'
[n other languages, including RUs,sian, the domain of a monomorphemic anaphor is the first finite
- '~' ( " J rl! ')
_;;
,
,
clause. In (67), below, we' see tRat the monomorphemic anaphor svoj 'self's' can be bound
. · ' ~ .;·f{Jii i .ni; .', , .
-:
outside an infinitival c1ause.'(68) shows that svoj cannot be bound outside of a finite clause «67)
: : ti~ : i )iL ';)r.,.'
~nrl (nR)
frnm
P.r()O()V~('
.:; " ,'
'
lQ(1)
cause. In t () /), oelOw, we
see tRat me monomorpnemlc anapnor SVOj selJ scan
. · '~ . .:'ffJVi .n{; .... ,' .
oe Duunu
1
outside an infinitival c1ause.'(68) shows that svoj cannot be bound outside of a finite clause «67)
::b "·~f ...tiL' ;~ r.l} ' ~;~".'
and (68) from Progovac, 1993)
,' .
(67)1=51
~. ff.d (j r1 ff
.
.
professori ~;' P9.~vS~s~ I...
assistentaj
[PROj citat' svoji/j dokladJ
professor
asked
assistant
read self's report
'The professoFris)W~ the:assisantj to read self'si/j report.'
,."
(67)[=12J
27
Vanjai znaet [eto
Vol odj aj
ljubit svoj~*i/j
zenu]
Vanja knows that
Volodja
loves self's'
wife
('Vanja knows that Volodja loves his own wife'') ' .'. . ,
.
Some researchers have proposed that the domain of monomorp~~.~ic: anaphors is closed off by
the first morphologically realized Agr. In Russian, this corresponds to the first finite clause; in
1
. '
",
f
J .
Chinese, there is effectively no limit on the domain of monomorphemic ·anaphors, because
-' .' .
,
'
Chinese has no morphological Agr (see. e.g., Progovac 1993) . .
Morphologically simple anaphors, as opposed to anaph~rs su?h as himself, are subjectoriented. That is, whenever a monomorphemi,c anaphor is boun~\ ?~~si,de of its local clause, it is
bound by a subject. In some languages, a monomorphemi¢ anap'lior is subject-oriented only
f
'.~,
.'
>
when it is long-distance bound. (68) and (69) below (from C6re': ~np~un$ 1994) show that Italian
proprio 'one's own' is subject-oriented when it is long-di'sfu~ce,'bound in (69), but not when it is
:
\ , ! , . H! "
locally bound in (68):
'I ".
';':.
-'
,
) '
(68)[= 12a]
recondotto
ha
Giannii
Maria"alla:!
':Propriail'
J .,
,," ~ ', \
J
Gianni
has
returned
Maria :to t!I~' ,; self.
'Gianni has brought back Maria to her own/his own family'
(69)[= 12b]
Osvaldo'
la
convinto
fatto che
ha
Giannii
.J,
,del
,':
Osvaldo . ,:, , of the fact
Gianni
convinced
t.lJ.at
t.lte
has
la
paese
casa e
del
pili .. bell~,'" "
propria *i/j
of the village
the
most beautiful
self
house IS
'Gianni convinced Osvaldo that his own house is 'nicest in the village.'
famiglia
family
.. ~ ~f ~· · <j~:,;,.t( ( ~';· " ~'
,,'..-
i
"
;
.
In other languages, including Russian, monomorphemic anapho'rs are subject-oriented even when
th;a t the Russian
they are locally bound. (70) below (from Progovac 1~), ~,hows
~
,;.?~ J '. . • (.
.
monomorphemic anaphor sebja 'self' is
•
.1;,
'
subject-oriented\~v'e~ 'w'p'~ji IH~:·locall·{boUlld.,;'
•
.:
•
A
,
j"'
(
"r .
' '' ~
1
,
,
milicioneri
rassprasival arestovanriogoj '
,: 0
se~ei/*j
policeman
questioned
suspect
~j .: : :;J,about self
'The policeman questioned the' susp'e'ct abodt hjmseli ..l· t,
.
j
(70)(=24J
.
, ..
j '•
•
I ' ,: .•
,.,
t.
• J
\~.~ ':I" ·\
.
•
,~~
In summary, morphologically complex anaphors ~r~ l~ca)~ ~~;~;~1r;~~d.:.allo~, antecedents of
-
r
~-- - .... ...... -
_.
,......... .
•
•
r.
1_ L _ I __ ~ __ 'If·_ :''' _ ':~_:''::'1 . 2 ;...~ ..... _L __
'The policeman questioned the' slIspe'ct abetlt himse lf ..ll
~.
: t ";~'-:':It: \.
• J
~
J
,._
I"'I
:
.....
t....o. _1_~.QC't
•
In summary, morphologically complex anaphors ~r~ l~ca)~ '~~.~J?~,,~?,dallo~, antecedents of
\. '''\: .''<'~ ' . .t lp.(~ ~:..,
•
." l,..
,
..
various grammatical functions. The domain "of mOrPh~I?~~;~H(~i,~R~e im?ph()~s is the closest
finite clause in Russian. and is unbounded in Chinese . .MoT"P,ho1~g~can y simple anaphors are
,
I, ~t),.(',) \ .. '. , "
•
subject-oriented when they are long-distance h?und. [n .S<?.m~.' la'ri$.aages, including Russian,
II
';
.f, <'l'tlLj ~U I!V
morphologically simple anaphors are subject-oriented "h.~i~;J~e~!~tj~i}o:~,ally bound as welL'.
.
~
,
'
.
28
5.2. Predicate Anaphora
Consider the similarities between the' b{'9dirig of the morphologically simple anaphors, as
..'
.
discussed above, and the case mar!Ging.6f exceptional secondary predicates.
i
The exceptional
second~rY
,preqicates:odin 'alone' and sam 'self,' are morphologically
simple. As s.ummarized in th~ tab~~"ii;n..(42) and (43), we saw that an exceptional secondary
.
i
.
no~n phrase in the nearest finite clause. Thus we have long-
predicate can agree in case with a
"
,
i , ..
....
"
distance cas.e agreement between a secondary predicate in an infinitival clause and a noun phrase
in the higher finite clause, as in' (~~>.;·,On~the other hand', an exceptional secondary predicate in a
subordinate finite clause agrees l",c~~JI!. ~i~ a local antecedent, as in (42c). We can therefore say
r
n
.
that the domain of case agr~ent f9.r:«r;~~eptional secondary predicates is the nearest finite
clause. This is identical to the 40m~ii;of.~inding for monomorphemic anaphors in Russian.
Furthermore, when an exceptional secondary predicate shows long-distance case
agreement,.it always agrees i~c~se
w,t~h
a subject; compare (42e), which shows that a secondary
predicate agrees in case with a'nqnloehl subject, to (42g), which shows that a secondary
r! tfllfll flWt' , . .
predicate does not agree in ~ase with a n~nlocal object. When an exceptional secondary predicate
r ..! ; . ' .j.
.
shows local case agreement, o,n. the other.hand, it is not subject oriented; (42a-b) show that an
_ oJ
i ...
.: -;~
!,
•
exceptional secondary predicate ag~s 'in case with its argument when that argument is in the
.-
'.
same clause, regardless of the
i j ~' i~ l.; :! ~,:
.
giall\m~tiq51rfunction
of that argument. Thus, the exceptional
secondary predicates are sUbject-:-priented for long-distance case agreement, but not for local
case-agreement. Note that thi~ pattern ,is ~i~!l~lT' to the subject-orientation of monomorphemic
anaphors in languages
suc::h:~..jl~~. Th~ '~~ssianmonomorphemic anaphors. svoj 'self's' and
.
sebja ' self, in contrast, are
alway~
'
~
'to. ,
I
subjecH>tiented .
• (\. 'J. ' ,l '
;
', : ' I t
>'
The similarities between,~Q.~ binai n~ of an'aphor,s and the case agreement of exceptional
.
' .'1 1:"" 1':' T ' 'I ; '
secondary predicates, as well a~ the' simat;l~~~ similarity between the anaphors and exceptional
A
,
.
.. .. .., i':t\"L,;;J
-, " I
I
•
~ \ L"
.I
"
: .,
-" .1
t l
.
"'.,,"'" '1'hesi~narii1es'bitwe~ru-~~~ bir;ding;oT~n'aPhors
secondary predicates, as well
.-J
1°
_
re
,..
and the case agreement of exceptional
L
•
~s}~h;!i~:;n~~nF similarity between the anaphors and exceptional
i" ":{ ,'.,}'I J: t - " .
secondary E~edicates. suggest that
exteptional secoo9ary predicates are, in fact, a type of
.
,
,: ••:J
t , ';;,' fj
: •
tne
•" ..,'.' . 1 (;(" :(! ...
I.' ..
·,
.
anaphor. ~t us call them "prerl:icat¢" a'naphqr~:' , ~~r convenience. We have already seen that
seconda~ pr~dicates can a~r~~11~'~~kc~ a ith, ~n' antecec;lent. Now we can make the generalization
. l :~t'lP)r£ '.' rr '-·; ' i
.
.
.
that exceptIOnal secondary predicates are bpund by an antecedent, and therefore agree m case
>I.
,~)! 51 fl V:)ri : ry"'~:"
.
with that antecedent. The "predicate anaphors" are monomorphemic. and therefore their domam
29
is the closest finite clause. Note that this conC\usion.f,u~ge,s~~ {~~~~~#~~ree of subject- ,,-,i!',' "
orientation, so to speak, is lexically determined. R,~s~~an, pr,~,di§\~t~,~?:~phors, in this an~ll~~is, are
subject-oriented only when they are long~distance bqu~~,;wpjl~8~~,~,~ ,Russian monomprph,emic
anaphors are always subject-oriented.
~"
c ~ -t . '!
((JrJ
Of course, we still need to account fophe occurre~~e}~! :~q~'i.Second Dative. Recall that
exceptional secondary pred~cates are accepted ~n em'Mon.m,~nt~j~M~hi~h they are not in ~ binding
relationship, either long,.di~nce or local, with a (fuqy
~ef.erem~~~b~tecedent.
.. .
we see that the argument;.o ( a secondary pr.~dicate can be f,ROl
obje~~~,Bu~ { note
or controlled by a direct or,Qblique
In (42d) and (42g)
'"
\
eve.~i:wh,en
PRO is uncontrolled
that,ill ~he~~,bO;.~iropments, the
.
~
.
.
secon~ary
.
predicate takes the dati~e case and does nO,t appear t~a~re~i,Il,w~,se ~~t~ any anteceden,t,: Ip terms
of a binding analysis, these non-agreeing s~c.opdary prepic'!.te~ }J,r~ l?~?phors.
Many researchers have 'pointed ~ut thflt anap,hors'.'l!..~ ,~O~~Pl~le,s acceptabl~ eV,e,n when
they are not bound, as in (71-72) below (~rQIIl Reinhart a~~ReN~~B2'J~~1).
(71)[=55a]
(72)[ =56a]
This paper was written by (Ann apd) myself. ; ,,, , , ,,,
Physicists like yourself are a godsend.
"'" A II, ,,,
.-
I
In examples such as (71-72), the anaphors are syntactically unbound, bunhe antecedent of each- '
anaphor can be inferred from the discourse context; these "anaphors" are called logophors.
Anaphors can be used as logophors in many languages, and Reinhart and Reuland (1991) note
that the distribution of monomorphemic anaphors used as logophors is especially free. It is not
.. ,
'
,"/' ,,' :',' ''1 -'
t',
r i' ,- , ;',. ,.
entirely unmotivated, then, to conclude that predicate anaphors c~n' b'e used as logophors
", p.,\
J
.-', :
'
whenever a binding relationship is not available. Since case agreement is not possible when a
1
.~
)1
~ ..
·· ~J :- F
~" "j d
.
,~ r
predicate anaphor is not bound by an antecedent, the predicate 'anaph6r arbitnirily takes the
JI
dative case.
~ ,
thf~" ~ri~lysls~:" r;it~t:~~b'~l(that long-distanc~ tase
agreement between a secondary predicate and a ~omr~a~'ive 's~b~~~llis' blocked by an overt ;
Two unresolved questions ';em'ain in
:.r . ~ } ~ c,-: .'
I
"P. ". / .. :f~.f
?trl l JO;
~" -~i
' d
.
~
•
complementizer. It is not clear why the presence of a complementlier shoilld 'affect bmdmg.
.
'"
_,1,
agreement between a secondary predicate and a ~omt~a~'ive 's~b~~~llis' blocked by an overt ;
"
:. r.
~l~
(: • .'
? . /
' f' ~~ ~ trl!
jn '
: _"- ~ '
'd '
.
complementizer. It is not clear why the presence of a 'com'pfementlier shobld'affect binding.
,
i i , ,, ~ .: j'! ; {~,'.; ~; ~ (. ,~-. ~ r.
,~
.~
Presumably, this question is related to the more general problem ofCdefin'tng the mechamsm for
the binding of secondary predicates.
Some researchers (e.g. Cole and
Ir
Sun~ 1994)"h~v~ it;gu11:t\handdg-distance binding is a
~ L'
'.',. ; \
~/ :
t' "
-'I"
-: (1 f·. ' ; J '~.
.
covertly local relationship, accomplished by head movemeniof'an' ahaphor to sutcessive verbal
30
projections at 'LE
argue
CleailYt~i~~JIiR'6t·~p'plY to'the binding of predicate anaphors. Even if we
tI~~{fu; single morp'hthies; p>rb'8tdie anaphors are heads and can move to I -
a dubious
, proposail.!(tlli~ ieads to diu'ti'desir~bte cOricfusion that case marking takes place at LF.
Other researchers (e.g. Progovac 1993) have argued that we should account for longdistance binding by param'etrlzibg tJie :cfoniain of binoing. That is, while morphologically
complex anaphors are 10call}Pt>bdnifth'iflbibCling of morphologically simple anaphors is truly a
lon~-distance relationship;:tilt JtYriia~'nippears to -d epend on the preserrce"of a morphologically
realized A'gr. Clearly, this' kind' of'ah~l)'si'S mllst also provide some way'to limit the possible
antec~dentS'of each anaphoEiProi6\1~~ (1993) argn~s that this can be accomplished by
parametrized the choice of-the aii~pn6r"rSub)ect based on whether the anaphor is a head or an
t
XP. Of course, it is pre~atutg~(nfrgO'~'; th~1t Progovitt's 1993) analysis can or cannot account for
the binding of predicate~iIaph~ts~;Ho'~ve'r, 'It is to be hoped that future work will provide a
binding mechanism that ;at~(jafits.f8t;t:h~1blnding of predicate anaphors as well as other anaphors,
and it appears that an account which irtt'olves lp arametrizing the binding domain could be
I
. "i.',
\t, ...
:
'-_~
6. Conc1us,i on
'JI-
;l:f~
>:
We have ~een that it is not ~~r~~~~tf,"~~~f~ to account for the case marking of exceptional
~.
.
.
.'
secondary, predicates in Russian. Accounts which assume that the case marking of secondary
(
,~ .
,
.
predicat~s i~ prod~~~\~J)!~gf}q2lf::h 5~~~}~~,~~{~r ~evis)~ons~o GB Theory are unsuccessful. We
saw that an account which requires PRO to be case-marked is m'echanically successful, but loses
a standard.generalizatiQn atwut the complementary distribution of PRO and lexical NPs. Finally,
~I
'
',
'
., '
,
J.
J _ , ~) :'; ~J1
J l-: 1'1
', j "
: .~
r:
t
\~;
..
~
10,
"I
I outlin,d a new analysi:sj ~n('~N:~~~~;~p'~i~n<~~f se:.o~~,ary predicates in Russian are taken to be a
kind oJ anaphor. The p,etaiis o(.this anal"y'sis
"
I
'
, I
;'
" .J :.
:. 5,\ n )r:L} r:l~ j ~L~',. : ,.1
r~main
{:
to be worked out. However, the analysis
'..<
I outlin,d a new '~alYSi:si~n('~R):~~~~;~mi~n<~}f ~~:.o~~,ary
pred-i~ai~~ 'in Russ1~~ -ar~-t.J(e~L·to- be a
kind 0ran~phor. Th~ Ae,tail~5~~J;~J~'lt?~)I~,~i~ te~ain to be worked out. However, the analysis
appears ~p. be quite p~()W.i' ~~~1Jtrg~~~~~es ! ~~: ~n.ter~~~ing generalization about the similarities
between the binding of morphologically complex anaphors and the case marking of exceptional
seconqary predicates. Furth~rmQre, the analysis suggests one way in which we may be able to
,~
. ,
" . c " .,~t!D(' JJ r~ ~ '\
4 ;·.',: l
.. ;:I;;:'
~' ,
accountJ9r,the _,~ase mark,~~~ 0Jl~~fR::~~~~~ ~r~condary predicates without introducing major
revisions to GB. It is to be hoped that future work will pursue this analysis in more detail.
References
Bachman, R. D. 1980. The Suhject Potential of the Dative Ca.~e in Modem RUlisian. Doctoml
dissertation, The Ohio Stilte University.
Bouchard, D. 1984. On the Conrent of Empty Categories. Dordrecht, Holland: Foris Publications.
189-193~
' .
'.
Cole, P., & Sung, L.M. 1994. "Head Movement and Long-Distance Reflexives," Lil1gui,litic
Inquiry 25: 355-406.
Comrie, B. 1974. '~he Second Dative: A Tran$formatiomil Approach," in R. D. Brecht & C. V.
Chvany (eds.), Slavic Tran.iformational Syntax; Ann ArbOrf MI: Michigan Slavic
Materials, No. 10. 123-150.
Davis, P. A., & Oprendek, D. V.I973. Making Progr'ess in Russian: New York, NY: John Wiley
& Sons.'
'
Franks, S. 1995. Parameters afSlavic Syntax. New York, NY: Oxford University Press.
Franks, S., & Greenherg, G. R~ J 988. "Agreement, Tense and the Ca~e of Subjects in Russian,"
in D. Brentari, G. Larson, & L. MacLeod (eds.); CLS 24: Papers from the 24th Annual
Regional Meeting ofthe Chicago Linguistic Society: Part Two:Paraliession on Agreement
in Grammatical Theory. 71-86.
.
Greenberg, G. R., & Franks, S. 1991. "A Parametric Approach·to-Dative Subjects and the Second
Dati ve in SI avi c:~Slavic and East European Jouma135: 7 J-97.
Napoli, D . .T. 1981. "Semantic Interpretation vs. Lexical Gpverilance: Clitic Climbing in Italian,"
Language 57: 841-887.
Preslar, R. M. 1995. The Syntax of Russian ImpersonaL Sentences. Doctoral dissertation,
University of Wa<;hington. Di.r;.r;ertationAhstracts Internationb155( 11): J496A.
Progovac, L. 1993. "Ling-Distance Reflexives: Movement-to-lnft ver~us Relativil.ed SUBJECT;:
Linguistic Inquiry 24: 755-772.
Reinhart, T., & Reiland, E. 1991. "Anaphors andLogophors: An Argument Structure
Perspective," in J. Koster and E.Reuland (eds.), [qng.Di.tlance Anaphora. Cambridge:
Cambridge University Press. 283-381.
.
Schei n, B. 1982. "Non-Finite Complements in Russian," MIT Working Papers in Linguistics 4:
217-244.
.
Sigurdsson, H. A. 1991. "Icelandic Case-Marked PRO and the I.icensing of I .exical Arguments,"
Natural Language and Lin~uistic Theory 9: 327-363.
.
217-244.
Sigurdsson, H. A. 1991. "Icelandic Case-Marked PRO and the f .icensing of I .exical Arguments,"
Natural Language and Linguistic Theory 9: 327-363.
.
Sigurdsson, H. A. 1992. "The Case of Quirky Subjects:' Working Papers in Scandinavian Syntax
49: 1-26.
'
Timberlake, A. 1986. ';The Semantic,:sof Case in Russian Predicate Complements," Russian
Linguistics 10: 137-165.
. .